Purpose:

WHole-genome Aligment Metricts (wham BAM) is a suite of tools designed to identify anomalies in binary alignment/mapping files (BAM).

The toolkit has two programs:

wham

Wham is designed to look for anomalies within a single bam file

whamy

Whamy is designed to look for anomalies across groups of bams.

Installing:

###Dependencies:

Cmake: Make your life easier and install Cmake. No action should be required under most linux distributions. If yes then use a package manager to install cmake.

Boost - At least 1.42: 1. Boot devel should be installed via Yum or other package manager. LibZ: 1. No action should be required under most linux distributions.

###Bamtools:

1. git clone https://github.com/pezmaster31/bamtools
2. cd bamtools
3. mkdir build
4. cd build
5. cmake ..
6. sudo make install

###wham BAM:

1. git clone git://github.com/jewmanchue/wham.git
2. cd wham
3. mkdir build
4. cd build
5. cmake ..
6. make
7. cd ..
8. rm -rf build

Testing:

sh /test/run-test.sh

Basic usage:

wham

wham mybam.bam mybam.bai scaffoldX > stdout 2> stderr

whamy

Whamy requires sorted BAMs. To improve whamy's speeds it is a good idea to remove duplicates and index the BAMS (not required). If the scaffold flag is not set whamy will run the whole genome.

whamy --target a.bam,b.bam,c.bam --background d.bam,e.bam,f.bam --scaffold chr1 > stdout 2> stderr

Output:

###wham

A tab delimited text file. Columns:

1. Seqid.
2. Position.
3. Number of "bad reads" covering the position in the pileup.
4. Number of reads covering the position in the pileup.
5. Fraction of "bad reads"
6. Probability of observing K bad reads out of N trials (See Details)
7. Global fraction of bad reads.

###whamy

A tab delimited text file. Columns:

1. Seqid.
2. Position.
3. Number of unmapped mates (target)
4. Number of unmapped mates (background)
5. Number of same strand mates (target)
6. Number of same strand mates (background)
7. Number of cross seqid mapped mates (target)
8. Number of cross seqid mapped mates (background)
9. Depth - Number of reads covering position (target)
10. Depth - Number of reads covering position (background)
11. Average insert length (target)
12. Average insert length (background)
13. Average mapping quality (target)
14. Average mapping quality (background)
15. Euclidean distances based on columns (3,4;5,6;7,8;13,14)

Details:

###wham

The Samtools flags are hashed across the entire scaffold. From this the global fraction of bad reads is calculated. Then the scaffold is "pileup." For each position with coverage the CDF of the binomial is calculated where:

K = number of bad reads (column 4) N = number of reads (column 5) P = global fraction of bad reads (column 7)

Bad reads include: Mate not mapped, and same strand. Further information will be added soon. IE: the insert size etc....